A tube’s measured body length and the length inferred from its resonance need not be identical. The air near an opening participates in the motion, and a real woodwind adds further acoustic structure. A ruler and a pitch measurement therefore answer related but different questions.
The gold standard of woodwind instruments in this lesson is to identify the acoustic model before using pitch to infer a length. This guide develops an original inverse problem: can several resonance measurements distinguish sound speed from an effective end correction?
Read The Gold Standard of Woodwind Instruments for the broader instrument system. The earlier clarinet register and written-pitch guide distinguishes musical and acoustic intervals. Here the question concerns what a resonance frequency can establish about an air column.
All tables and numerical examples are invented for learning. They describe simplified uniform tubes with stipulated boundary conditions. They are not measurements of a particular flute, clarinet or other instrument, and they do not give instructions to modify one.
eduKateSG · Woodwind Instruments
Find your next learning step
Use reciprocal-frequency plots to infer effective length corrections, test acoustic assumptions and distinguish length changes from sound-speed changes.
Chapter contents
Understand the evidence · Chapters 1–4
Work the model · Chapters 5–8
Test assumptions · Chapters 9–12
Practise and repair · Chapters 13–16
Report and continue · Chapters 17–20
CHAPTER 1 OF 20 · Understand the evidence
1. Define physical length and effective length
Let L be the physical length measured between the endpoints specified in a tube model. Let δ be the model’s total effective length correction, and define Leff = L + δ. The correction represents how the acoustic boundary differs from an ideal boundary placed exactly at the physical endpoint.
In a simplified open–open cylindrical tube, the lowest resonance is approximately f = c/[2(L + δ)], where c is sound speed. For an idealised open–closed tube, the corresponding expression is f = c/[4(L + δ)]. The boundary condition changes the denominator.
The same symbol δ does not imply the same correction for those two systems. It is the total correction appropriate to the stated model. Two open ends and one open end do not automatically contribute an identical total.
Begin a solution by writing the length’s endpoints and boundary conditions in words. “The instrument is 60 centimetres long” is not yet enough information. It may describe an external dimension, a path along the body, or an acoustic segment associated with a particular fingering.
CHAPTER 2 OF 20 · Understand the evidence
2. Explain the open boundary without adding imaginary tubing
At an ideal open end, the elementary standing-wave picture places a pressure node at the boundary. A real opening interacts with the surrounding air, so the effective acoustic boundary is displaced from the simplest geometric endpoint.
The UNSW woodwind acoustics explanation describes radiation impedance and an approximate open-end correction for a simple cylindrical pipe. It also explains why real instruments introduce further complications. That account supports the physical motivation for using an effective length.
An effective correction is not a hidden piece of solid tube. It is part of an acoustic model. Extending a ruler beyond the opening may help visualise the idea, but the surrounding air does not become a rigid cylinder with a perfectly fixed end.
For the calculations below, δ is a stipulated constant for a family of otherwise comparable idealised configurations. This assumption is deliberately testable. If the fitted correction changes strongly with frequency, geometry or configuration, the constant-correction model may no longer describe the observations adequately.
CHAPTER 3 OF 20 · Understand the evidence
3. Calculate the forward prediction first
Assume an open–open model with c = 340 m/s, physical length L = 0.480 m and total correction δ = 0.020 m. The effective length is 0.500 m, giving f = 340/(2 × 0.500) = 340 Hz.
If the correction were omitted, the predicted frequency would be 340/(2 × 0.480), approximately 354.167 Hz. The uncorrected prediction is about 4.167% higher than the corrected value. The comparison uses the corrected value as its percentage reference.
The direction is useful before the calculation. Adding a positive correction increases effective length, which lowers the predicted fundamental at fixed sound speed. If a solution predicts a higher frequency after adding a positive correction, inspect the formula or units.
This forward case is the foundation for the inverse problem. A learner who cannot predict frequency from a fully specified model will struggle to infer hidden parameters from a measurement. Practise one forward substitution, one unit check and one direction check before reversing the relationship.
CHAPTER 4 OF 20 · Understand the evidence
4. Show why one frequency does not identify two parameters
Suppose the supplied physical length is 0.480 m and the observed fundamental is stipulated to be 340 Hz. Under the open–open model, the equation is c = 680(0.480 + δ).
One possible pair is c = 340 m/s and δ = 0.020 m. Another is c = 333.2 m/s and δ = 0.010 m. Both satisfy exactly the same single observation. The pitch alone does not select between them when both c and δ are unknown.
This is an identifiability problem rather than an arithmetic failure. Recalculating the same equation more precisely does not create a second independent relation. More decimal places on the frequency cannot by themselves distinguish the two parameters.
An independent estimate of sound speed would permit a model-based correction estimate. Alternatively, several measurements under conditions that share the same c and δ can provide additional relations. The next chapters show the second route and explain the assumptions that make it legitimate.
Starting from f = c/[2(L + δ)], take the reciprocal to obtain 1/f = (2/c)L + 2δ/c. Define y = 1/f and x = L. Then y = mx + q, where m = 2/c and q = 2δ/c.
The slope has units of seconds per metre. Its reciprocal relation gives c = 2/m. The intercept has units of seconds, and δ = q/m gives a length in metres. These unit checks help prevent confusing the intercept itself with a length correction.
A plot of frequency against physical length is curved in this model. A plot of reciprocal frequency against physical length is straight when c and δ remain constant. Linearisation is useful because it makes the two parameters visible as slope and intercept.
It does not make the assumptions true. A straight-looking line over a narrow range may be compatible with several models. A strong explanation names the physical conditions held constant and then uses additional measurements to check whether the transformed relation remains adequate.
Take two supplied configurations from the same idealised family: L1 = 0.480 m with f1 = 340 Hz, and L2 = 0.660 m with f2 = 250 Hz. Assume the same sound speed and total correction in both.
The reciprocal-frequency difference is 1/250 − 1/340, approximately 0.001058824 seconds. Divide by the length difference 0.180 m to obtain slope m = 1/170 seconds per metre, approximately 0.005882353 s/m.
Therefore c = 2/m = 340 m/s. The intercept is q = 1/340 − m × 0.480 = 1/8,500 seconds. Dividing q by m gives δ = 0.020 m.
| Physical length L | Effective length in the stipulated model | Fundamental frequency | Reciprocal frequency |
|---|---|---|---|
| 0.480 m | 0.500 m | 340 Hz | 0.002941176 s |
| 0.660 m | 0.680 m | 250 Hz | 0.004000000 s |
| 0.830 m | 0.850 m | 200 Hz | 0.005000000 s |
The third row has not been used to fit the two parameters. It provides a predicted value that can be checked independently, provided its configuration belongs to the same model family.
CHAPTER 7 OF 20 · Work the model
7. Check a third configuration rather than refitting immediately
Using c = 340 m/s and δ = 0.020 m, the prediction at L = 0.830 m is f = 340/[2 × 0.850] = 200 Hz. If the supplied third observation is 200 Hz within its stated uncertainty, it is consistent with the fitted model.
If the third observation is instead 204 Hz, the inferred effective length using c = 340 m/s would be 340/(2 × 204), approximately 0.833333 m. Relative to the stated physical length, the implied correction is about 0.003333 m.
That differs from the fitted 0.020 m. Possible explanations include changed sound speed, a configuration-dependent correction, an incorrect mode label or measurement error. The discrepancy does not identify which explanation is correct.
Record the prediction before examining the third result. This makes the check meaningful and prevents a later fit from being described as an independent prediction. If a revised model is introduced, retain both the earlier prediction and the evidence that motivated the revision.
CHAPTER 8 OF 20 · Work the model
8. Do not assume every fingering shares one correction
Changing a woodwind fingering changes more than a single geometric endpoint. Tone-hole geometry, the remaining tube beyond an opening and interactions among openings can affect the acoustic response. A constant δ across every note is therefore a teaching simplification, not an established instrument law.
The two-length exercise is most useful when learners can name why the configurations are treated as comparable. In supplied data, that may be an explicit assumption. In an investigation, it requires a defensible experimental design and an assessment of how the relevant geometry changes.
A visually straight reciprocal-frequency plot does not prove that all openings act identically. Measurement range and uncertainty may conceal smaller departures. Conversely, a departure from the simple line can be educational evidence of the model’s limits rather than evidence that the instrument is defective.
Connect this question to The Gold Standard of Acoustics. The broader subject explains how sources, boundaries and receivers contribute to a measurement. Here the inverse problem shows why a measured pitch cannot be interpreted independently of that physical system.
CHAPTER 9 OF 20 · Test assumptions
9. Separate a sound-speed change from a length change
At fixed effective length, the elementary model predicts frequency proportional to sound speed. At fixed sound speed, frequency is inversely proportional to effective length. A changed frequency therefore does not uniquely identify a changed length.
For a supplied comparison, let effective length remain 0.500 m while c changes from 340 to 346 m/s. The predicted fundamental changes from 340 to 346 Hz. The ratio is 346/340, approximately 1.017647, an increase of about 1.765%.
If an analyst incorrectly holds c at 340 m/s, the new 346 Hz observation implies Leff = 340/692, approximately 0.491329 m. The calculation would suggest a shorter effective length even though the construction explicitly held length fixed.
This counterexample demonstrates the need for controlled conditions or independent sound-speed evidence. It does not claim that a real instrument’s response to temperature is captured entirely by this one equation. Actual playing conditions can change several parts of the system together.
The fundamental formula applies to the lowest mode of the stated idealisation. If the selected spectral peak belongs to a higher mode, using the fundamental relation gives a misleading length even when the measured frequency is accurate.
For an ideal open–open tube with c = 340 m/s and Leff = 0.500 m, the first two mode frequencies are 340 and 680 Hz. Substituting 680 Hz into the fundamental relation gives an inferred length of 0.250 m, half the stipulated effective length.
The error comes from the mode label. A strong spectral peak is not automatically the fundamental. The excitation and measurement arrangement can make some components more prominent than others, so amplitude alone is not a complete mode-identification rule.
Ask the learner to explain the expected sequence of modes and the evidence used to label the observed peak. In a real woodwind, resonance peaks and played sound components also need careful distinction. The simplified tube exercise provides a reasoning framework, rather than a universal spectrum-reading shortcut.
For the ideal open–closed fundamental, f = c/[4(L + δ)]. Its reciprocal form is 1/f = (4/c)L + 4δ/c. The slope is now 4/c, while the ratio of intercept to slope remains δ.
This creates an instructive ambiguity. A reciprocal-frequency line alone, without the boundary-condition model, does not establish which factor belongs in the sound-speed calculation. The same measured slope would imply different c values under open–open and open–closed assumptions.
The physical boundary description and mode pattern provide information that the line fit does not contain. Fitting a straight line is not a substitute for deciding which acoustic system is being modelled.
As an original check, take c = 340 m/s, L = 0.480 m and δ = 0.020 m in an open–closed model. The fundamental is 170 Hz. The result is half the open–open prediction for the same stipulated effective length. Explain that the boundary condition changed before comparing the numbers.
CHAPTER 12 OF 20 · Test assumptions
12. Keep correction estimates within their evidence
If c is independently supplied, the open–open inverse is δ = c/(2f) − L. Suppose c = 340 m/s and L = 0.480 m, while f is known only to lie between 338 and 342 Hz.
Because c/(2f) decreases as f increases, the smallest correction comes from 342 Hz: approximately 0.017076 m. The largest comes from 338 Hz: approximately 0.022959 m. The correction therefore lies between about 17.1 and 23.0 millimetres under the stated fixed-c model.
This interval accounts only for the supplied frequency bound. If c and L are uncertain, their contributions must also be considered. A small difference between two estimated corrections may not be meaningful when their uncertainty ranges overlap substantially.
The decimal places in a calculator result do not establish measurement resolution. Report the assumptions alongside the estimate. If the inverse calculation produces a physically surprising correction, first inspect boundary conditions, mode labels, units and the adequacy of the simplified model.
CHAPTER 13 OF 20 · Practise and repair
13. Explain why a near-zero intercept can be fragile
The fitted correction is δ = q/m. When the intercept q is small compared with uncertainties in the measured reciprocal frequencies, modest measurement changes can alter the estimated correction substantially even if the slope remains fairly stable.
The problem becomes visible when all physical lengths are clustered tightly. The fit must infer where the line would cross the vertical axis far from the observed range. A tiny slope change can move that intercept appreciably.
For teaching, compare two exact datasets first, then introduce small perturbations and observe how the parameters respond. Do not treat the exact two-point arithmetic as a full statistical uncertainty analysis. Real measurements may need an appropriate regression method and a treatment of uncertainty in both axes.
A good report distinguishes “the model fits these two points exactly” from “the correction is known accurately”. The first follows algebraically from two parameters and two observations. The second requires information about measurement uncertainty, model adequacy and independent checks.
CHAPTER 14 OF 20 · Practise and repair
14. Diagnose the learner's first unstable step
If a learner confuses centimetres and metres, return to unit conversion before fitting a line. A length of 48 centimetres is 0.480 metres. Combining 48 with a sound speed in metres per second changes the scale dramatically.
If the forward formula is secure but the inverse problem is not, present two parameter pairs that reproduce one frequency. Ask what extra observation would distinguish them. This exposes the difference between solving an equation and identifying several unknowns.
If the learner fits the line correctly but calls its intercept “two centimetres”, ask for the intercept’s units. Since the vertical axis is 1/f, the intercept is a time. Dividing by the slope recovers a length.
For extension, change the boundary condition or mode label and ask which part of the calculation must change. A learner who can identify the affected assumption is better prepared for unfamiliar problems than one who can only repeat the original substitutions.
CHAPTER 15 OF 20 · Practise and repair
15. Practise with answers and changed conditions
Task A: an open–open model has c = 330 m/s, physical length 0.530 m and total correction 0.020 m. Find the fundamental. Answer: Leff = 0.550 m and f = 330/1.100 = 300 Hz.
Task B: a second configuration in the same stipulated family has physical length 0.640 m. Predict its fundamental. Answer: effective length is 0.660 m and f = 250 Hz. The two reciprocal-frequency points give c = 330 m/s and δ = 0.020 m.
Task C: one measured frequency and one physical length are available, but both c and δ are unknown. Is there a unique solution? Answer: generally no. State the relation between c and δ and identify the missing independent evidence.
Task D: a peak at twice the expected fundamental is used in the fundamental inverse formula. What happens in the ideal open–open example? Answer: the inferred effective length is halved. The mode label must be checked.
Task E: does a positive correction raise frequency at fixed physical length and sound speed? Answer: no. It increases the denominator and lowers the predicted frequency. Explain the direction before calculating.
In the first stage, label the physical model. Draw the tube, specify its boundary conditions and mark the measured length. Use a supplied sound speed and one clear fundamental frequency. The goal is meaningful substitution.
In the second stage, reverse the formula with only one unknown. Infer effective length from c and f, then compare it with physical length. State which values were supplied and which were inferred.
In the third stage, use two configurations to estimate c and δ together. Transform to reciprocal frequency, track the units of slope and intercept, and check both original observations by substitution.
In the fourth stage, introduce a third observation, an uncertain frequency or a changed mode. Ask the learner to decide whether the old model still answers the question. This progression connects calculation with evidence rather than treating a more difficult formula as the only sign of progress.
A parent can ask, “What did we measure, and what did we infer?” A teacher can ask, “Which assumption lets these two configurations share one correction?” Both questions direct attention to the structure of the reasoning.
Is the correction always the same for every note? No. The constant correction is an assumption in the worked family. Real geometry and frequency-dependent effects can change the effective boundary behaviour.
Can the outside length of a flute determine its lowest note exactly? Not through this simple model alone. The relevant acoustic geometry, openings, sound speed and boundary effects must be understood.
Does a good line fit prove that the model is physically complete? No. It shows agreement over the tested observations within the stated analysis. Additional checks can reveal departures or competing explanations.
Should a learner alter an instrument to test the formula? The supplied-data exercises require no instrument modification. They teach inverse reasoning using declared models and original numbers.
Why does the guide use simple tubes? They make the relation between boundary conditions, effective length and frequency inspectable. Their usefulness comes from both what they explain and the limits they make visible.
CHAPTER 18 OF 20 · Report and continue
18. Compare a constant correction with a proportional correction
The straight-line construction assumed a constant total correction δ. Consider an alternative teaching model in which the effective length is (1 + k)L, with k a constant proportion. Then f = c/[2(1 + k)L], so 1/f = [2(1 + k)/c]L.
This alternative gives a line through the origin. Its slope contains the ratio (1 + k)/c. Without independent information, the line does not separately identify both k and c. More length measurements can refine the slope while leaving that parameter ambiguity intact.
For example, c = 340 m/s and k = 0.02 produce the same slope as c = 333.333333… m/s and k = 0 under this construction. The first has a proportional correction and the second has a lower stipulated sound speed. Their reciprocal-frequency relation is identical.
This does not imply that the proportional model is a better model of a real instrument. It is a counterexample showing why a successful transformed fit must be interpreted through its physical assumptions. The equation that defines the parameters determines what the data can identify.
Ask learners to compare three statements: “the points lie close to a line”, “the constant-correction model describes the tested range” and “the physical correction is definitely constant”. Each statement makes a stronger claim than the previous one. Additional evidence is needed before moving from a convenient mathematical representation to a broad physical conclusion.
An independent sound-speed estimate or a controlled geometry comparison can help discriminate between competing parameter explanations. The choice of additional observation should follow the ambiguity in the model, rather than simply collecting more of the same measurement without a clear purpose.
CHAPTER 19 OF 20 · Report and continue
19. Trace how a small frequency change affects inferred length
With fixed c, the open–open inverse length is Leff = c/(2f). Suppose a frequency changes from f to f + Δf. The exact ratio of new inferred length to old inferred length is f/(f + Δf). This is more precise than saying the percentage changes are always exactly equal and opposite.
For f = 340 Hz and Δf = 3.4 Hz, the frequency increases by one percent. The inferred length is multiplied by 340/343.4 = 1/1.01, approximately 0.990099. It decreases by approximately 0.9901%, rather than exactly one percent.
For sufficiently small changes, the equal-and-opposite percentage rule is a useful approximation. The exact ratio shows both why it works near zero change and why it should not be treated as an identity for large changes.
The inferred correction can be relatively more sensitive than the effective length because δ = Leff − L subtracts two quantities. In the main example, Leff is 0.500 m while δ is only 0.020 m. A change of 0.005 m is one percent of Leff but twenty-five percent of δ.
This explains why a pitch-based effective-length estimate can appear stable while a small inferred correction varies substantially. The correction is a difference, and its uncertainty depends on both the inferred effective length and measured physical length.
For a practice extension, retain c = 340 m/s and L = 0.480 m, then compare inferred corrections at 340 and 343.4 Hz. The effective lengths are 0.500 m and approximately 0.49505 m; corrections are 0.020 m and approximately 0.01505 m. Ask which percentage comparison is meaningful for the question being answered.
CHAPTER 20 OF 20 · Report and continue
20. Report the inference and follow the connected guides
A useful report states the tube idealisation, mode, length endpoints, shared-parameter assumptions, fitted slope and intercept, recovered parameters and independent check. It labels the invented data as a teaching construction.
For the main example: “Under a shared open–open model, the 0.480 m/340 Hz and 0.660 m/250 Hz configurations give c = 340 m/s and total correction δ = 0.020 m. The model predicts 200 Hz at physical length 0.830 m. This prediction tests the model only if the third configuration shares the stated assumptions.”
Continue through Woodwind Instruments, Acoustics and Resonance. These guides connect the calculation with sources, boundaries, modes and musical use.
The primary physical reference is UNSW: the acoustics of woodwind musical instruments. Its discussion motivates effective boundaries and the need for care with real instruments. The inverse datasets, fitted parameters, uncertainty exercise and teaching progression above are original explanatory work.
Contents · Previous chapter · Return to Woodwind Instruments